Not a technology list. The stuff underneath it.

Every tool on the projects page started somewhere. Usually it’s one of these, recognised in the wild before it ever became a spec.

Users rarely ask for the thing they actually need.

• The spreadsheet that became load-bearing

  A tracker that started as one person’s personal list and quietly became the system three teams now depend on, that nobody’s allowed to touch.

  Why I like it — under the conditional formatting is usually a real, decent data model. It just needs a proper home.

• Systems that won’t talk to each other

  Two platforms that each do their own job well and still can’t agree on a customer record, so someone re-keys it by hand every week.

  Why I like it — the interesting part is never the API. It’s working out which system should actually own the truth.

• The report that answers the wrong question

  Four hours a month to build, and it answers a question nobody actually asked, twice removed from the decision it was meant to inform.

  Why I like it — half the job is technical. The other half is sitting with someone until they say what they actually meant.

• The process that lives in one person’s head

  It only works because one person remembers every exception, and everything stops the week they’re on leave.

  Why I like it — that’s not a training problem. That’s an undocumented system wearing a person as its interface.

• AI without a problem attached

  Being asked to “add AI” to something before anyone’s agreed what it’s actually meant to fix.

  Why I like it — the honest answer is usually a smaller, more boring automation. Less exciting in a meeting. Right anyway.

• The bottleneck everyone’s routed around

  A step so slow the whole team has quietly built a workaround, and the workaround is now the actual process.

  Why I like it — workarounds are free user research. Someone already told you exactly where it hurts.